New Physical Evidence of the Role of Stream Capture in Active Retreat of the Blue Ridge Escarpment, Southern Appalachians

Geomorphology 123 (2010) 305–319

Contents lists available at ScienceDirect

Geomorphology

journal homepage: www.elsevier.com/locate/geomorph

New physical evidence of the role of stream capture in active retreat of the Blue Ridge escarpment, southern Appalachians

Philip S. Prince ⁎, James A. Spotila, William S. Henika

Department of Geosciences, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA article info abstract

Article history: The Blue Ridge escarpment of the southern Appalachian Mountains is a striking and rugged topographic feature of Received 8 February 2010 the ancient passive margin of eastern North America. The crest of the escarpment generally coincides with an Received in revised form 13 July 2010 asymmetric regional drainage divide, separating steep streams of the escarpment face from low-gradient streams Accepted 26 July 2010 of the Blue Ridge Upland. Recent exhumation and erosion rate studies suggest that the escarpment has evolved by Available online 11 August 2010 inland erosional retreat, but the mechanism, timing, and magnitude of retreat remain poorly understood. Longitudinal stream profiles and slope–drainage area relationships of several upland basins draining the divide Keywords: fi fl Appalachians have led to the identi cation of 14 previously unknown uvial terrace deposits preserved at the escarpment crest. Landscape evolution These relict terraces and the associated beheaded drainages indicate the role of large stream capture events in Escarpment retreat producing ongoing escarpment retreat through landward divide migration and subsequent topographic Passive margin escarpment adjustment. Terrace location and preservation suggest that rectilinear drainage patterns and divide asymmetry Stream capture generate discrete high order captures and episodes of rapid localized retreat that collectively produce slower Divide migration evolution of the escarpment at large. While overall retreat magnitude and rate remain unknown, roundness of terrace alluvium suggests that the most recent captures have locally produced tens of kilometers of retreat within the limited preservation lifetime of the deposits. In contrast with recent numerical modeling and cosmogenic studies, these data show the potential for stream capture and divide migration to sustain passive margin escarpment evolution long after the cessation of rifting. The fluvial record of divide retreat preserved atop the Blue Ridge escarpment suggests the potential for using field methods to better constrain the histories of younger, taller, and potentially more dynamic passive margin escarpments. © 2010 Elsevier B.V. All rights reserved.

1. Introduction al., 1994; Steckler and Omar, 1994; Brown et al., 2000; Spotila et al., 2004). The existence of a universal paradigm applicable to great Major seaward-facing topographic escarpments are common fea- escarpment evolution is thus questionable. tures of rifted passive margins. Such “great” escarpments are believed to Further constraining the chronology and mechanisms of great initiate during rift-flank uplift and subsequently evolve through escarpment retreat, particularly along mature passive margins, is erosional processes (Ollier, 1984; Kooi and Beaumont, 1994; Gallagher significant to improving our understanding of the evolution of passive and Brown, 1997). Rapid erosion focused on the steep flank of a rift may margin landscapes. After rifting, escarpments may be rapidly excavated produce inland retreat (Ollier, 1984; ten Brink and Stern, 1992; Young by downwearing seaward of a fixed drainage divide (van der Beek and and McDougall, 1993; Tucker and Slingerland, 1994; Seidl et al., 1996), Braun, 1999; Matmon et al., 2002) or experience slow but steady retreat but the controls on the rate, timing, and mechanism of escarpment in parallel with the divide (King, 1962; Fig. 1). However, no clear global migration following rifting are poorly understood. Some great escarp- relationship exists between escarpment age (determined by age of ments, such as southwestern Africa and southeastern Australia, appear rifting) and distance from the coast; the Sri Lanka escarpment (180 Ma, to have migrated little since the early stages of development (Moore et 65 km from the present coastline; Vanacker et al., 2007), the Drakens- al., 1986; Gilchrist et al., 1994; Bishop and Goldrick, 2000; Cockburn et berg escarpment (130 Ma, 150 km from the present coastline; Moore al., 2000; Matmon et al., 2002; Persano et al., 2002), while other and Blenkinsop, 2006), and the southeastern Australia escarpment escarpments, such as southeastern Brazil and the Blue Ridge escarpment (85–100 Ma, 60 km from the coastline; Weissel and Seidl, 1998)imply of eastern North America, may have continued to retreat into the late different overall retreat distances and long-term average rates. The Blue stages of escarpment development (Bohannon et al., 1989; Gallagher et Ridge escarpment (BRE), a passive margin escarpment often overlooked in great escarpment studies, occurs along the oldest passive margin in the world (initial rifting at ~200 Ma; Pique and Laville, 1995; McHone, ⁎ Corresponding author. Tel.: +1 864 320 3243; fax: +1 540 231 3386. 1996), yet it is located within 70 km of inland rift basins and maintains E-mail address: [email protected] (P.S. Prince). steep, youthful topography (Spotila et al., 2004). The timing of

(A) escarpments exist as topographic steps separating low-elevation, Preexisting inland divide low-relief coastal plains from elevated continental interiors. Great escarpments are typically ~0.3–1 km high and exist as narrow bands (~5–20 km) of rugged topography separating the lower relief coastal plain and upland surfaces. They are generally classified as either Rift “shoulder-type” or “arch-type,” depending on the position of the (B) regional divide in relation to the crest of the escarpment (Matmon et Initial divide al., 2002). The hypothesis of escarpment evolution by landward retreat after rifting is based on observations of active rift margins, where dynamic continental uplift and base level drop seaward of the divide produce focused erosion and steep, seaward-facing landforms. Rift Recent studies of thermochronometry, cosmogenic radionuclides, (C) and offshore sedimentary data have suggested some escarpments Initial divide evolve by initially rapid erosion that slows considerably as seaward- flowing streams begin to adjust to rift-related base level (Drakensberg escarpment (Fleming et al., 1999), Namibia (Rust and Summerfield, 1990; Cockburn et al., 2000; Bierman and Caffee, 2001), Sri Lanka (von Blanckenburg et al., 2004; Vanacker et al., 2007), and southeastern Rift Australia (van der Beek and Braun, 1999; Persano et al., 2002)). Continued knickpoint retreat on these seaward-flowing rivers may, Fig. 1. Comparison of three popular models of passive margin escarpment evolution. Dotted lines trace the development of escarpment topography across equal time however, increase escarpment sinuosity over time with little retreat intervals. All models ultimately produce escarpments of similar morphology despite of the topographic front (Weissel and Seidl, 1998; Heimsath et al., differing mechanisms and rates of retreat. (A) Rift-related base level drop seaward of a 2000; Matmon et al., 2002). In contrast, other data suggest significant preexisting drainage divide steepens large streams and rapidly excavates an escarpment retreat continues long after rifting. Inferred exhumation escarpment at the divide. Subsequent retreat is minimal. (B) Rift flank uplift produces an asymmetric divide atop the rift shoulder. Focused erosion on the steep flank of the histories of escarpments surrounding the Atlantic basin, such as divide produces steady parallel retreat of divide and escarpment. (C) The parallel eastern Brazil, southwest Africa, and eastern North America are retreat mechanism of B decelerates because of the decrease in escarpment relief. consistent with continued late-stage retreat (Gallagher et al., 1994; After van der Beek et al., 2002. Brown et al., 2000; Spotila et al., 2004). Drainage patterns along the Western Ghats escarpment suggest continuing capture of upland exhumation of the area seaward of the BRE is comparable with that streams, leading to overall parallel retreat of the drainage divide and observed seaward of the ~100 Myr younger southeastern Australia escarpment (Harbor and Gunnell, 2007). escarpment, each with low temperature (U–Th)/He apatite cooling ages Numerical modeling has shown the influence of pre-rift topogra- of ~85–110 Ma (Persano et al., 2002; Spotila et al., 2004). The persistent phy, particularly the location of the regional drainage divide, on the ruggedness of the BRE, despite its age, makes it unique among great rate and mechanism of escarpment evolution (Fig. 1). Modeling by escarpments, and further constraining the mechanism and potentially van der Beek and Braun (1999) and van der Beek et al. (2002) showed the timing of retreat is necessary to understanding long-term BRE that initial rapid evolution and subsequent stability are consistent evolution. The results of this study may, in turn, find application to other with downwearing from a preexisting inland divide (arch-type of great escarpments, where establishing additional controls over long- Matmon et al., 2002). Large streams flowing seaward from the divide term retreat mechanisms may enhance the interpretation of existing are energized by rift-induced base level drop and rapidly excavate an cosmogenic and thermochronologic data sets. escarpment at the divide as they equilibrate to the new base level. In this paper, we examine fluvial terraces preserved at the crest of Following this equilibration, the lower-energy seaward stream head- the BRE and physical characteristics of beheaded upland drainage basins waters accomplish minimal additional retreat (Fig. 1A). The more as a possible record of ongoing parallel escarpment retreat. Terraces traditional model of prolonged, steady escarpment retreat involves were systematically located through the analysis of upland drainage the formation of a new, asymmetric regional divide, which retreats in basin morphology. Rounded to well-rounded alluvium from the terraces parallel with the escarpment as a result of constant, focused erosion was subsequently sampled for clast roundness measurements, which on the steep flank and upland stability (King, 1962)(Fig. 1B). were used to estimate clast transport distance. The use of field-based While cosmogenic data indicating very slow retreat of many mature data aids in testing recent models for the origin and evolution of the BRE escarpments are consistent with the downwearing model (e.g., generated by apatite thermochronometry (Spotila et al., 2004), Cockburn et al., 2000; Brown et al., 2002), slow escarpment erosion cosmogenic radionuclides (Sullivan et al., 2007), and drainage basin could result from reduced relief and thus be consistent with the parallel geometry (Bank, 2001). Terraces provide new field evidence of retreat model as well. Climatic effects have also been suggested as a significant parallel retreat of the escarpment and divide continuing means of accelerating or slowing parallel retreat (Partridge, 1998). throughout the Cenozoic. Analysis of beheaded drainages provides an Thermochronological data do not clearly distinguish between down- additional indicator of escarpment retreat and drainage divide wearing or parallel retreat because of low geothermal gradients and encroachment. The preservation and predictable location of surficial modest escarpment relief of ~1 km or less (Braun and van der Beek, and topographic evidence of BRE retreat also suggests the possibility of 2004). The contrast between these models makes constraining retreat using field techniques to better characterize the evolution of younger, mechanism an important step toward describing the origin and and potentially better preserved, passive margin escarpments. evolution of escarpments. Downwearing and parallel retreat scenarios have been modeled to ultimately produce escarpments of similar morphology despite different rates (Fig. 1), suggesting that the present 2. Background morphology of escarpments may not provide clues to the mechanism of their evolution (Braun, 2006). The application of field methods aimed at 2.1. Long-term evolution of great escarpments determining the role of the drainage divide may add a useful additional constraint to the modeling of escarpment evolution. Clearly identifiable great escarpments occur along approximately Understanding escarpment retreat mechanisms has equally one-third of the world's passive margins (Ollier, 1984). These profound implications for the maintenance of rugged landforms P.S. Prince et al. / Geomorphology 123 (2010) 305–319 307 over time. The relief-reducing processes of Davis (1902) and Penck southwest from Roanoke, Virginia and is characterized by rolling (1953) explain low-relief upland and lowland surfaces, but fail to hills, thick soils and saprolites, and aggraded valleys (Fig. 2). Average account for passive margin escarpment steepness. Parallel slope slope atop the Upland in Virginia is ~10°, but the surface becomes retreat as explained by King (1957) allows for the maintenance of increasingly rugged to the southwest (Spotila et al., 2004). The Upland relief over time, as the high potential energy of steep hillslopes extends ~40 km from the crest of the BRE northwest to the Valley and focuses erosion while allowing upland areas to remain comparatively Ridge fold-thrust belt (Figs. 2, 3A). Northeast of Roanoke, the Blue stable. This concept is now viewed as largely dependent upon Ridge highlands rise above the Piedmont as a narrow ridge isolated lithology and structure, such as resistant caprocks and escarpment- from the highlands farther west. This ridge morphology appears to parallel jointing that stabilize uplands and ease removal of material have developed as a result of Atlantic basin rivers (e.g., Roanoke, from the escarpment face (Munro-Perry, 1990; Moore and Blenkin- James, and Potomac) breaching the western margin of the Upland and sop, 2006; Gunnell and Harbor, 2008). In contrast, Penck (1953) rapidly eroding the weak sedimentary rocks of the eastern Valley and hypothesized that uplands should erode along with the steeper Ridge (Harbor, 1996). In contrast, the Blue Ridge Upland forms a escarpment face, reducing relief over time. This long-standing debate plateau that remains physically and fluvially connected to the of landform evolution may find some resolution in the study of old westward-draining areas of the Valley and Ridge and Appalachian passive margin escarpments. Plateau to the north and west. Several large westward-flowing streams with headwaters on the Upland surface ultimately drain 2.2. The Blue Ridge escarpment portions of all three provinces (Fig. 2). As a shoulder-type escarpment, the majority of the crest of the BRE The BRE stretches over 500 km along the eastern edge of the coincides with the Eastern Continental Divide (ECD), an asymmetric southern Appalachian highlands (Fig. 2). Rising in 300–600 m of relief regional drainage divide between Gulf of Mexico (west-flowing) and above the 200–300 m Piedmont, the BRE represents an abrupt Atlantic (east-flowing) streams (Hayes and Campbell, 1894; Davis, topographic boundary between the lower-relief Piedmont and Blue 1903; Wright, 1927; Dietrich, 1957; Hack, 1973)(Figs. 2, 3). The low Ridge Upland (or simply, the “Upland”) surfaces. Long wavelength gradient of the westward-flowing streams draining the Upland slopes average 24° in the BRE zone in Virginia, with steeper slopes contrasts strongly with the steep headwaters of streams draining developed on granitoid rocks in North Carolina (Spotila et al., 2004). the narrow BRE zone (Fig. 4). The steep escarpment slopes provide In contrast to most high-relief zones in the Appalachians, the steep energy to Atlantic basin streams plunging toward the Piedmont topography of the BRE does not coincide with the outcrop of a whose erosive power is evident in cascading bedrock reaches and resistant lithology. Bedrock units on either side of the escarpment oversteepened fluvial profiles. Hack (1973) viewed the atypical offer essentially equal resistance to erosion, and contacts between steepness of the escarpment slope streams compared to Upland and units cut across the BRE at varying angles. The BRE also shows no clear nearby Piedmont streams draining the same lithology as evidence of relationship to the nearby Brevard/Bowens Creek fault zone (Fig. 3), fluvial disequilibrium along the BRE. These streams flatten rapidly which crosses the BRE in western North Carolina (Fig. 3)(Dietrich, upon reaching the low-relief Piedmont where generally symmetrical 1959; Hack, 1973; Virginia Division of Mineral Resources, 2003). drainage divides and lower stream gradients suggest a more Landward of the BRE, the Blue Ridge Upland forms a broad plateau, equilibrated landscape. Some Atlantic basin streams draining the termed the Floyd Surface by Dietrich (1957), which extends escarpment appear to have captured Upland streams, such as the Dan

Fig. 2. Location of the Blue Ridge escarpment and Eastern Continental Divide (ECD) in relation to relevant topographic and geologic features of the southern Appalachians. The escarpment is the area of steep topography separating the Blue Ridge (BR) and Piedmont physiographic provinces. South of Roanoke, Virginia (R), the escarpment coincides with the ECD (yellow line). Mesozoic basins in the Piedmont are shown in green. Brevard/Bowens Creek fault zone (BFZ) is identiﬁed by red line. Fall line (FL) is denoted by gray line. VR = Valley and Ridge, AP = Appalachian Plateau, and GSM = Great Smoky Mountains. Boxes denote locations of Fig. 3A and B. 308 P.S. Prince et al. / Geomorphology 123 (2010) 305–319 P.S. Prince et al. / Geomorphology 123 (2010) 305–319 309

Fig. 4. Long proﬁles of three pairs of Upland and Atlantic basin streams (Fig. 3A) illustrating the asymmetry of the Eastern Continental Divide. The overall convex proﬁle of the Ararat River (ii) indicates a large zone of active incision that may result from a recent capture event that stranded terrace E (Fig. 3A).

River of Virginia (stream 19, Fig. 3A), whose headwaters meander Sparse field evidence of parallel retreat, in the form of well- across the Upland before turning 90° and dropping over 600 m to the rounded alluvium preserved near the BRE and divide and the character Piedmont (Dietrich, 1957; Hack, 1973). Apparent redistribution of fish of beheaded streams, has previously been reported (Wright, 1927; species of the New River basin into streams comprising the head- Dietrich, 1957, 1959; Bank, 2001). Rounded cobbles and boulders waters of the Atlantic-draining Roanoke River basin (streams 19, i and scattered in dry valleys or on hilltops near the divide are believed to ii of Fig. 3) provides further qualitative evidence of the capture of have been stranded by a reduction in drainage basin area and westward-flowing drainages (for a review, see Jenkins et al., 1971). competence of westward-flowing Upland streams through landward These topographic and biological data provide anecdotal evidence of escarpment and divide migration (Dietrich, 1957, 1959). This stream capture and episodic divide and BRE retreat. proposed origin of the alluvium is consistent with qualitative The BRE has long been viewed as a feature shaped by erosional observations suggesting drainage basin reduction in westward- retreat related to the asymmetry of the divide (Davis, 1903; Wright, flowing Upland streams that now originate at the ECD. Many Upland 1927; White, 1950; Dietrich, 1957, 1959; Harbor, 1996) but has seldom streams begin to meander very close to their headwaters at the been regarded as a great escarpment produced by rift-flank uplift (e.g., escarpment lip, while Piedmont streams of similar size show no Ollier, 1984; Tucker and Slingerland, 1994). Pazzaglia and Gardner meanders (Spotila et al., 2004). Wide, flat valleys and swamps are (2000) suggested that the BRE is a great escarpment that was excavated common at headwaters in close proximity to the divide, atypical of the from a fixed inland divide after Mesozoic rifting and base level drop. headwaters of other streams atop the Upland surface. Quantitative Other hypotheses for its origin have favored active tectonic or study of fluvial profiles and slope–drainage area relationships of these geodynamic origins. White (1950) proposed that normal-sense reacti- streams may aid in verifying basin reduction as a product of divide and vation of the nearby Brevard/Bowens Creek fault zone produced the escarpment retreat. Establishing the provenance of escarpment-crest escarpment. Numerous workers have cited isostatic rebound related to alluvium should further enhance understanding of retreat kinematics. thickened Appalachian crust or flexural response to offshore sediment The use of field-based proxies for escarpment evolution has proved loading as drivers of BRE formation and evolution (Wright, 1927; Pratt somewhat useful in other locations, such as localized deposits et al., 1988; Battiau-Queney, 1989; Hubbard et al., 1991; Pazzaglia and preserved seaward of the Western Ghats escarpment (Widdowson Brandon, 1996; Pazzaglia and Gardner, 2000). Spotila et al. (2004) and Gunnell, 1999; Harbor and Gunnell, 2007), weathering surfaces presented thermochronological constraints on BRE evolution. Cooling and duricrusts of the South African upland (Partridge and Maud, histories of the Upland and Piedmont are distinct, suggesting that they 1987), and the offshore stratigraphic record of erosion along the BRE do not represent a single offset “surface.” These data also showed no (Poag and Sevon, 1989; Naeser et al., 2006) and southwestern Africa difference in cooling ages across the Brevard/Bowens Creek fault zone, (Rust and Summerfield, 1990). Alluvium preserved at the escarpment suggesting that reactivation of the structure did not form the BRE. The crest and evidence of drainage basin reduction may thus directly greater exhumation that has occurred on the Piedmont during the past reflect the history of divide retreat and provide concrete physical 100 Ma is consistent with the erosional removal of a topographic bulge constraints on the evolution of the BRE. seaward of the present escarpment and, hence, long-lived retreat. While these data argue against local faulting or uplift as the origin of the BRE, 3. Methods they do not distinguish between escarpment formation through differential erosion by parallel retreat or downwearing from a To obtain empirical constraints on the kinematics of BRE retreat, preexisting or fixed inland divide. Thermochronology also fails to we have investigated the morphology of beheaded streams on the indicate distance and timing of retreat or association with an initiating Upland and characterized associated relict alluvium preserved atop structure in the present Piedmont. Verifying the mechanism of erosional the divide. Both of these sources of data should reflect any loss of basin retreat and constraining the kinematics of retreat are essential to testing area from parallel retreat of the divide and escarpment. Progressive the varied hypotheses of BRE origin. encroachment of the divide into established Upland basins may have

Fig. 3. (A) Shaded relief map of Virginia study area showing streams (numbers) and terrace deposits (letters) used. Larger numbers indicate large streams whose profiles are presented in Fig. 5B. P and P′ are endpoints along a topographic profile of the ECD (Fig. 8). ECD and Brevard fault zone (BFZ) are indicated. Map is based on 10-m resolution DEM. Streams are labeled as follows: 1 = Little River, 2 = Payne Creek, 3 = Silverleaf Branch, 4 = Meadow Creek, 5 = Thomas Grove Church tributary to Dodd Creek, 6 = W. Fork Dodd Creek, 7 =Oldfield Creek, 8 = W. Fork Little River, 9 = Spurlock Creek, 10 = Big Indian Creek, 11 = Rock Creek, 12 = Greasy Creek, 13 = Burks Fork tributary, 14 = Burks Fork, 15 = Adams Branch, 16 = Chisholm Creek, 17 = Laurel Fork, 18 = Bear Creek, 19 = Dan River (Atlantic basin), 20 = Stone Mountain Branch, 21 = Big Reed Island Creek headwaters, 22 = Pine Creek, 23 = Puckett Church Branch, 24 = Pipestem Branch, 25 = Sulphur Springs Branch, 26 = Grassy Creek, 27 = Grassy Creek tributary, 28 = Pine Creek Ward's Gap, 29 = Little Snake Creek, 30 = Big Reed Island Creek, 31 = S. Prong Buckhorn Creek, 32 = Little Reed Island Creek, 33 = E. Fork Crooked Creek, 34 = Crooked Creek, 35 = Chestnut Creek, 38 = Brush Creek, i = Rennett Bag Creek, ii = Ararat River headwaters, iii = Waterfall Branch. (B) Shaded relief map of North Carolina study area showing streams and terrace deposits used. Map is based on 10-m resolution DEM. 37 = Terry Creek, 38 = Flat Creek. Terraces J, L, and M and Terry Creek are drained to the Atlantic Ocean by the Green River, a stream analogous to the Dan River (19) of the Virginia study area. 310 P.S. Prince et al. / Geomorphology 123 (2010) 305–319 removed steep headwater reaches of fluvial profiles, leaving behind shape were selected to quantify an Upland control stream slope–area the lower gradient of downstream morphologies. By comparing these relationship describing streams unaffected by divide retreat. This potentially beheaded drainage profiles to unaltered streams atop the slope–area relationship was combined with the average length–area Upland, estimating the magnitude of basin area or stream length lost power law relationship from the same ten streams to obtain an may be possible. Alluvium of obvious fluvial origin preserved at the expression for length (L) as a function of slope (S): headwaters of beheaded streams may exceed the competence of the host stream given present relief and discharge and be excessively −1:68 rounded for present channel length. Qualitative comparison of clast L = ðÞS=0:0335 ð2Þ roundness to an established relationship to transport distance may provide an estimate of the magnitude of divide and escarpment migration. Together, these data may confirm the mechanism of BRE This expression describes the relationship between increasing retreat and permit a generalized restoration of the BRE through channel length and decreasing local slope of Upland control streams paleobasin reconstruction. unaffected by divide retreat. Slopes at or near the headwaters of many Graphical and quantitative methods can be applied to the ECD-draining streams are anomalously low and more consistent with description of stream channel and basin geometry. A smooth, slopes observed well downstream of typical Upland stream head- fi concave-up longitudinal profile is regarded as an indicator of steady- waters. We substituted the unusually low initial slope values (de ned fi fi state equilibrium in a fluvial system (Hack, 1957, 1973; Snyder et al., by the length of the rst contour interval in the stream) of the ve 2000; Roe et al., 2002; Whipple, 2004; Bowman et al., 2007; Goldrick most linear streams draining the asymmetric divide for S in Eq. (2) to and Bishop, 2007; LaRue, 2008), with disruptions in smooth concave- estimate channel length lost to divide and BRE retreat. up shape indicating disequilibrium, such as that produced by faulting To constrain retreat using relict alluvium preserved at the escarp- fi or basin capture. The loss of headwaters to divide migration could be ment crest, we rst completed a systematic search for terrace deposits at manifest as linear profiles whose steepened headwaters, and thus the headwaters of ECD-draining streams that appear truncated by the concave-up shape, have been lost to headward erosion of Atlantic divide. We focused on streams originating in broad gaps or low-relief drainages. Longitudinal profiles provide a qualitative basis of compar- areas at the escarpment crest. The gentle gradient of these headwaters ison, whereas the slope–drainage area relationships of streams can be areas, characteristic of downstream reaches of Upland control streams quantitatively compared. Local channel slope and drainage area are landward of the escarpment and divide, presented qualitative evidence related by the power law of basin loss and offered the greatest potential to preserve relict alluvium at the surface. Field reconnaissance was conducted at the −θ headwaters of all ECD-draining streams whose morphologies indicated S = kA ð1Þ beheading to locate physical evidence of basin loss from divide retreat, where S is local slope, k is the steepness index, A is drainage area such as mature alluvium preserved in wind gaps very near or atop the upstream of the point of the slope measurement, and θ is the ECD. Four of the headwaters terraces that hosted the most abundant and concavity index (Flint, 1974). Values of θ should be generally best-preserved rounded clasts were selected for transport distance consistent within tectonically and lithologically similar regions estimates based on clast roundness. fl (Kirby and Whipple, 2001; Whipple, 2004). Beheaded streams lacking Clast roundness is known to increase with progressive uvial a concave-up profile shape should present less negative θ values and transport such that the shape of clasts in alluvium preserved at the lower slope at a given drainage area (or stream length) than unaltered divide may be inverted to obtain the magnitude of channel length lost Upland streams of comparable size. to BRE and divide retreat (Mills, 1979; Lindsey et al., 2007). Sadler and fi We constructed longitudinal profiles and slope–area relationships Reeder (1983) offered an empirical and eld-expedient method for based on 1:24,000-scale USGS topographic maps for 20 westward- relating the roundness of quartzite clasts to transport distance. Their flowing streams with headwaters at the ECD (the ECD-draining regression was produced in the San Bernardino Mountains, California, streams) and 18 westward-flowing control streams from the Upland using clasts of clearly discernible provenance, permitting a direct “ ” interior (the Upland control streams). ECD-draining profiles were first outcrop-to-basin determination of transport distance that could be compared to Upland control profiles to identify characteristics related to roundness. We chose the method and regression of Sadler consistent with divide migration and basin loss, such as low-gradient and Reeder (1983) because it was developed for a similar lithology headwaters and overall linear (not concave-up) shape. Average and was produced empirically using clasts of known origin and Upland control and ECD-draining profiles were produced by averaging transport distance. Their results showed good agreement with a fl slopes from all streams in each population in 1000-m increments. The quartzite clast ume experiment of Kuenen (1956) as well as other qualitative data gathered from profile analysis was then used to guide empirical studies (Tricart and Schaeffer, 1950; Hovermann and Poser, our selection of streams for slope–area analysis. When the effects of 1951; Hollerman, 1971; Goede, 1975). Clast shape is expressed divide and escarpment retreat on the drainage pattern of the study quantitatively by the Cailleux Roundness Index, or CRI (Cailleux, fi area are considered, it is apparent that not every stream rising at the 1947), de ned as divide and flowing west would have been beheaded. During landward migration, the major regional divide would occasionally “overtake” = CRI =2*fgðÞRc L *1000 ð3Þ preexisting subordinate divides once located in the Upland interior. Headwaters of streams rising at these subordinate divides would then rise at the ECD and escarpment crest, but would not have been affected where Rc is the radius of curvature of the sharpest corner and L is the by beheading. A random selection of ECD-draining streams would mix length of the long axis. To avoid low estimates for thinly bedded these equilibrium streams with beheaded streams, potentially quartzites or tabular veins, Rc is measured in the orientation of obscuring evidence of divide and escarpment retreat. Accordingly, maximum projection (parallel to short axis, orthogonal to long and we applied an intentional bias to focus attention on ECD-draining intermediate axes). After Sadler and Reeder (1983),weonly streams that have most likely been beheaded. We chose the five most measured clasts with long axes ranging from 4 to 10 cm to avoid linear and apparently anomalous streams draining the ECD westward transport over-estimates related to the greater ease of rounding of to estimate stream length lost through comparison to streams draining very large clasts. We measured CRI for ~30–60 of the most rounded the same surface but rising landward of the ECD. Ten Upland control unbroken clasts present in each terrace to obtain a maximum estimate streams whose profiles showed smooth, well-developed concave-up of transport distance (i.e., using an intentional bias). P.S. Prince et al. / Geomorphology 123 (2010) 305–319 311

Fig. 5. (A) Long profiles of short (≤6 km) ECD-draining streams (left) and Upland control streams (right) (Fig. 3A, B). Average profiles were developed by averaging slope values over 1000 m intervals for every stream within each of the four groups. Locations shown in Fig. 3. Asterisks denote streams with terrace deposits at their headwaters. (B) Long profiles of ~25-km-long ECD-draining streams (left) and Upland control streams (right) (Fig. 3). Little Snake Creek (29; right) drains a high elevation, high-relief area that was likely a local divide within the Upland, and is therefore treated as an Upland control stream despite its present proximity to the ECD. Average profiles constructed in the same manner as A.

4. Results profiles are nearly linear and lack the steep headwater reaches that produce the concave-up shape of Upland control profiles (Fig. 5). The Longitudinal profiles indicate morphological contrasts between more pronounced concave-up shape of the average Upland control ECD-draining and Upland control streams. A number of ECD-draining profile implies fluvial equilibrium with lithology and regional base 312 P.S. Prince et al. / Geomorphology 123 (2010) 305–319 level, whereas the comparatively linear shape of the average ECD- Knickpoints, or local convexities, occur along profiles of both draining profile suggests the loss of headwaters to fluvial beheading. populations (e.g., streams 10 and 31 of the control group; 5 and 34 of This trend in profile shape is apparent in both short (~6 km) and long ECD-draining streams), but do not alter the overall shape of profiles (~25 km) streams (Fig. 5). (Fig. 5). These may represent active rejuvenation of a relict westward- flowing drainage network by intermittent lowering of the landward base level. Slight contrasts in lithologic resistance may also produce the isolated convexities, but the effects are localized and are not mapped (Virginia Division of Mineral Resources, 2003). The unusually steep profile of stream 11 likely results from the resistant Precam- brian–Cambrian quartzites underlying its basin; it was excluded from average Upland profile calculation (Fig. 5). Several ECD-draining streams show concave-up profiles or exhibit slightly steepened headwater reaches (3, 6, 23, 33) (Fig. 5). These profiles may represent preexisting equilibrium streams of the Upland interior that were “overtaken” in place by landward divide migration and thus have not lost drainage area. Alternatively, this morphology may be the result of an earlier pulse of New River incision which migrated headwardly through the drainage network and rejuvenated the basins to their headwaters at the ECD. Aside from these variations, however, a clear distinction between the two stream populations is evident in the individual and averaged profile shapes (Fig. 5). A comparison of the steepness and concavity indices of streams draining the asymmetric divide to the Upland control streams further delineates morphological distinctions between the two groups and provides an estimate of channel length lost to divide retreat. We focused our analysis on five streams rising at the divide that presented particularly linear profiles and low headwater slopes: 4, 22, 35, 36, and 38 (Figs. 3, 5). ECD-draining streams show a more gradual downstream decay of channel gradient, resulting in less negative concavity index (θ) values. Average concavity index (θ) for the ECD-draining streams is −0.219, compared to −0.401 for the ten Upland control streams analyzed. Morphological distinctions between the two stream populations are also apparent in steepness index (k; Eq. (2)) values; average steepness index (k) for ECD-draining streams is 0.157, while the steeper headwater reaches of the Upland control streams yield an average (k) value of 0.401. These slope–area data reflect the trend observed in the longitudinal profiles and are consistent with basin loss resulting from divide and escarpment retreat. Substituting the low headwater slopes of the five ECD-draining streams into Eq. (2) provides a loose estimate of stream length lost to divide retreat. Estimated channel loss results are presented in Fig. 6. Average estimated channel loss from BRE and divide retreat is ~8 km (Fig. 6). Chestnut Creek (35) and Flat Creek (38) suggest the largest channel losses (18.0 and 11.1 km, respectively). While loose estimates, these analyses of channel concavity appear to reflect significant channel loss and are thus consistent with divide retreat and sufficient stability of the Upland to preserve the relict basin morphology. Combined analysis of longitudinal profiles, slope–area relationships, and 1:24,000-scale topography led to the identification of 14 individual fluvial terrace deposits in wind gaps and low-relief areas atop or near the ECD (Fig. 3). These terraces are physical evidence of

Fig. 6. (A) Slope–drainage area relationships of ten Upland control streams showing plotted raw data and best-fit power law expressions. Average Upland control slope– area relationship is described by the power law S=0.0314A−0.403. Ellipse 1 indicates the range of Upland control stream headwater slopes; ellipse 2 indicates range of headwater slopes of beheaded ECD-draining streams (C). (B) Drainage area-channel length relationships of the ten Upland control streams from A. Average relationship is described by the power law A=0.855L1.47. Average slope–area (A) and area–length (B) power laws were combined to obtain the average Upland control slope–length power law (L=(S/0.0335)−1.68) plotted in C. (C) Headwater slope values of five beheaded ECD-draining streams shifted to fit the average Upland control slope–length power law curve. ECD-draining headwater slopes are anomalously low and consistent with slopes encountered kilometers downstream from Upland control stream headwaters. Red arrows indicate the magnitude of the required shift and thus the Upland control channel length necessary to obtain the low headwater slopes of the ECD-draining streams. P.S. Prince et al. / Geomorphology 123 (2010) 305–319 313 both parallel retreat of the assymetric divide and stability of the weathering of the deposit complicates this interpretation. Clasts are Upland surface. The terraces are characterized by accumulations of very abundant in the deposits, a number of which approach a clast- rounded to well-rounded vein quartz or quartzite clasts up to ~25 cm supported structure. All clasts show signs of weathering, but the long (Fig. 7). No polymineralic (lithic) clasts were observed in any of extent of weathering is variable between deposits located in close the terrace deposits. Fine matrix material typically is sandy clay; but proximity to one another. For example, site E contains numerous some deposits, particularly site N (Figs. 3B, 7), contain a matrix clasts showing little evidence of chemical weathering; but site I, dominated by quartz sand and refractory minerals. Clay content and located ~15 km away (Fig. 3A), contains only extensively weathered redness appear to be proportional to the extent of clast weathering. and pitted cobbles, most of which are broken. The absence of a trend The effects of floroturbation (i.e., tree throw) and agricultural between clast weathering or soil development and terrace location disturbance are apparent (Fig. 7). Terrace I (Fig. 3A) may contain suggests ongoing, but episodic, asymmetric divide retreat. primary depositional features indicated by roughly parallel lines of The locations of the terrace deposits are related to the local cobbles ~1 m below the surface in a roadcut (Fig. 7), but the extensive topography. Deposits concentrate in wind gaps and low-relief areas

(A) Terrace E (B) Terrace N

Terrace E Terrace B

133 km

(F) 10 cm (E) Terrace I 279 km

Fig. 7. Selected images of terraces showing typical ﬁeld characteristics. All photos are taken within ~100 m of the ECD. Locations of terraces are shown in Fig. 3. Scale card arrow is 10 cm long. (A) Large, well-rounded clasts in sandy clay matrix from terrace E. (B) Unweathered vein quartz clasts in quartz sand matrix from terrace N. (C) Rounded to well- rounded clasts exposed by ﬂoroturbation (tree throw) from terrace B. (D) Small boulder from terrace E. (E) Cobble lines exposed by roadcut from terrace I. (F) Selected clasts from terrace E. Note rounding at all clast sizes. 314 P.S. Prince et al. / Geomorphology 123 (2010) 305–319

1400 SW NE 1200 1000 G 800 I H F E(D) C B A 600

Elevation (m) Elevation 400 Vertical exaggeration 7x 0 10 20 30 40 50 60 70 80 90 P Distance (km) P’

Fig. 8. Profile along the ECD from P to P′ (Fig. 3A) showing terrace locations. Terraces cluster in low-elevation, low-relief areas along the ECD. within broad, region-scale sags along the divide (Fig. 8). In the stream length by divide and BRE retreat. We measured clast roundness Virginia study area (Fig. 3A), these sags cluster at 825 m (~2700 ft) for ~30–60 clasts from four terraces (N, E, B, A) (Fig. 9). Roundness was and 760 m (~2500 ft) elevation. Many terraces occur just at or above converted to an estimate of transport distance using the relationship of the headwaters of streams whose profiles show little or no concave- Sadler and Reeder (1983). Terraces N and E show the highest roundness up shape, low headwater gradient, and a lack of rejuvenation from and estimated transport distances for individual clasts, with a maximum landward base level drop (Fig. 5). Preserved unconsolidated alluvium value of ~270 km (clast shown in Fig. 7F). Average transport distance is not, however, found at the headwaters of all such streams or in all and filtered average (excluding highest and lowest 10% of values) are sags along the asymmetric divide. Several preserved terraces are lower, but all suggest stream length loss of N10 km (Fig. 9). Given that located seaward of the divide (e.g., D, J, L, M; Fig. 3). These terraces not all clasts within a terrace would have been transported the same share the same elevations and topographic features as other sites on distance, the average roundness values of the top 10%, ranging between the Upland margin and seem to be perched on topographic remnants 47–185 km, may be more useful metrics. Some of the variation between of a once-continuous Upland surface. For example, terrace D appears terraces may relate to “paleoorder” or paleoflow direction of the streams to align with stream 24 and is preserved at the same elevation as that deposited the terraces. The two terraces with lower average terrace E, possibly reflecting divide retreat through two distinct transport values (A, B) occur in tributaries of the same master stream, episodes of stream capture (the headwaters of stream 30 were theLittleRiver(Figs. 3A, 9). In contrast, the more rounded clasts of captured to form stream 19, followed by additional capture of stream terraces E and N occur in valleys that are directly contiguous with 30 headwaters by stream ii; Fig. 3). The preservation of unconsoli- escarpment-orthogonal trunk streams. dated alluvium atop these small Upland remnants separated by deep, narrow gorges suggests that sudden, rapid base level drop resulting 5. Discussion from stream capture can produce strongly differential erosion in mature landscapes. The apparent stability of these Upland remnants Anomalous drainage basin morphologies and fluvial terraces provide adjacent to active gorge development invites comparison to the strong qualitative evidence for ongoing divide and escarpment retreat contrast in Upland and stream valley erosion rates at Dolly Sods, West resulting from the repeated capture and dissection of Upland drainage Virginia, described by Hancock and Kirwan (2007). basins. These data suggest a model of escarpment evolution that differs Roundness of the clasts preserved in remnant terraces at the crest of from the recent preexisting, or fixed, inland divide plateau degradation the BRE imply significant transport and hence considerable loss of model derived from cosmogenic, thermochronologic, and numerical

300 300 Terrace N Terrace E

250 250 n=41 90 -100% n=58 filtered mean (10-90%)=61.4 km filtered mean (10-90%)=28.0 km 200 200 mean (90-100%)=185.9 km mean (90-100%)=149.0 km 90-100% 150 150

100 100 10-90%

50 50 10-90%

Estimated transport distance (km) 0 0

60 60 Terrace A 90-100% Terrace B 50 50 n=25 n=26 90-100% filtered mean (10-90%)=16.1 km filtered mean (10-90%)=18.9 km 40 40 mean (90-100%)=51.4 km mean (90-100%)=42.8 km 30 30 10-90% 20 10-90% 20 distance (km) 10 10

Estimated transport 0 0

Fig. 9. Distribution of estimated clast transport distances from terraces N, E, B, and A (Fig. 3). Transport estimates are based on measured values of Cailleux Roundness Index from each clast (Cailleux, 1947) after Sadler and Reeder (1983). P.S. Prince et al. / Geomorphology 123 (2010) 305–319 315 modeling studies of other escarpments (Moore et al., 1986; Gilchrist et its role in escarpment retreat. Terraces A and B have lower average al., 1994; Bishop and Goldrick, 2000; Cockburn et al., 2000; Matmon et roundness values and are drained by the BRE-parallel local trunk al., 2002; Persano et al., 2002; van der Beek et al., 2002; Braun and van stream, the Little River (stream 1; Fig. 3A), and may thus be the der Beek, 2004). The morphology of Upland drainages rising at the remnants of short, lower order tributary streams only truncated by divide is consistent with a state of transient adjustment between the the last few kilometers of asymmetric divide retreat. Terrace E is Upland and BRE (Dietrich, 1957; Hack, 1973; Gasparini et al., 2007). drained by a stream directionally and topographically continuous Examples of topographic disequilibrium observed cannot be rectified with the local trunk stream, Big Reed Island Creek (stream 30; with a near-stationary BRE that stopped retreating significantly long ago Fig. 3A), which flows roughly orthogonal to the BRE. The greater (Sullivan et al., 2007). We instead propose a BRE history similar to the roundness of terrace E clasts may reflect longer transport in the larger structurally and fluvially controlled retreat model suggested for the basin of a paleo-trunk stream. Clasts from terrace N, which is also Western Ghats escarpment by Harbor and Gunnell (2007) and Gunnell drained by a stream showing general continuity with the French and Harbor (2008). We view BRE evolution as an episodic, yet ongoing, Broad River (Fig. 3B), provided the highest overall transport estimates process where significant long-term parallel retreat is accomplished from all terraces studied. Terraces N and E, which host the most through the repeated capture and rapid dissection of Upland drainage rounded alluvium, also showed the least amount of weathering, basins. suggesting relatively recent stranding and thus rapid erosional While the overall distance of escarpment migration remains destruction of their depositing basin upon its capture. This compar- uncertain, altered basin morphologies and clast roundness allow for atively fresh alluvium may be viewed as additional qualitative loose estimates of recent (106 years) retreat. The vein quartz clasts evidence of the rapid dissection and local divide and escarpment measured here certainly equal, and likely exceed, the resistance to retreat associated with large capture events. rounding of the quartzite clasts of Sadler and Reeder (1983), We propose a conceptual model of BRE evolution that highlights suggesting that our transport estimates (i.e. ~50–100 km based on the role of stream capture in producing escarpment retreat through maximum roundness) are probably conservative. Structural control of prolonged, but punctuated, retreat of the ECD (Fig. 10). Headwardly drainage immediately southeast of the BRE by northwest–southeast eroding Atlantic basin streams of the steep divide flank (Fig. 10, panel trending fractures or joints suggests that tens of kilometers of 1) can tap the potential energy of westward-flowing Upland transport could have been orthogonal to the escarpment, suggesting drainages through stream capture (panel 2). The capture process transport distance could provide a reasonable basis for a loose retreat may initiate with diversion of Upland groundwater before the estimate. Headwater gradient of channels truncated by the divide westward-flowing surface channel is physically diverted to the suggests ~10–20 km of channel loss, reasonably consistent with Atlantic basin. Connection to the seaward base level, which is averaged clast transport estimates. Flow direction and meander hundreds of meters lower, greatly steepens and energizes the interval are, however, significant unconstrained variables; and the captured stream, and rapid incision propagates headwardly through escarpment-parallel flow and high degree of sinuosity present in the basin as it equilibrates to the new base level (panel 2). Structural streams of the study areas would contribute to an overestimate of weaknesses (faults and fractures/joints) controlling flow directions retreat. Clast transport distance will thus remain a rough estimate of will facilitate rapid dissection of the captured basin and allow retreat magnitude until sediment provenance is clearly established. headward erosion to encroach upon neighboring Upland basins, The systematic preservation of terrace deposits at low-gradient eventually producing additional captures along strike (panel 3). headwaters supports the conclusion that apparent channel loss is the Adjustment of captured streams to the seaward base level rapidly result of divide and escarpment retreat. The persistence of surficial dissects the detached Upland remnants and moves the locus of alluvial deposits and anomalously low headwater slopes at the topography landward in response to the new location of the divide. escarpment crest suggests Upland denudation is outpaced by Continued headward erosion along structural weaknesses sets up landward retreat of the divide and escarpment. Although the terraces future landward captures and re-starts the retreat process (panel 4). are not dated, clast weathering is similar to Upland-sourced vein In order for this retreat model to continue in the long term, capture quartz clasts from N1 Ma New River terraces preserved under the and retreat must occur more rapidly than lowering of the Upland same climate conditions in the nearby Valley and Ridge (Ward et al., (landward) base level to maintain divide asymmetry and the 2005). While a potential age of more than 1 Ma for our deposits energetic potential of Upland streams relative to the seaward base suggests great stability of the surface on which they are preserved, level. Minor (tens of meters) episodic incision of Upland streams due Upland denudation rates of ~10 m/Myr based on thermochronometry to landward base level drop could potentially propagate headwardly (Spotila et al., 2004) and cosmogenic dating (e.g., Sullivan et al., 2007) through the entire drainage network and ultimately steepen the suggest a limited lifetime of surficial deposits atop the Upland. If the headwaters of westward-flowing Upland streams, increasing sym- terraces are several million years old and were stranded by the metry of the divide. A topographic ridge may form along the ECD, capture of escarpment-orthogonal drainages (terraces D and E; acting as a physical impediment to future capture events. Over the Fig. 3A) tens of kilometers in length, local capture-driven retreat long term, the combined effects of repeated Upland base level drops rates of 1–10 km/Ma are plausible. This implies that local divide and will reduce contrast in landward and seaward base levels and thus BRE retreat rates may, in the case of major capture events, outpace reduce the potential energy of Upland streams available for capture Upland lowering by as much as three orders of magnitude. These (bottom of panel 4). Divide asymmetry and retreat thus form a speculative retreat rates would, however, only result from the positive feedback loop; asymmetry facilitates rapid retreat, which in occasional capture of large Upland drainage basins with discharges turn maintains the divide morphology and the energetic potential for capable of producing very rapid erosion upon capture. Indeed, smaller continued capture and retreat. A sufficiently stable landward Upland but more frequent captures producing comparatively modest local base level is thus equally vital to preserving retreat potential in the retreat rates probably account for considerable retreat over the long long term and allowing the process to continue after lengthy intervals term. In either case, BRE and divide retreat rates exceeding Upland between localized captures. lowering by an order of magnitude or more would be limited to While we view BRE evolution as a long-lived and ongoing process, captured basins undergoing dissection and not be applicable to the the local, basin-scale retreat rates of ~1–10 km/Ma proposed are entire BRE at any given time. certainly not applicable to the whole feature for the entirety of its When considered in the context of the present Upland drainage inferred ~200 Ma lifetime. Cosmogenic dating indicates the BRE is no network, the respective location, roundness, and freshness of terrace longer experiencing significant retreat along much of its length (e.g., alluvium may offer some insight into the paleodrainage network and Sullivan et al., 2007). This suggests rate-controlling events must be 316 P.S. Prince et al. / Geomorphology 123 (2010) 305–319

Terrace

1 2

Terrace Former escarpment Terrace position (1)

Terrace 4 3

Fig. 10. Illustration of the conceptual model of parallel divide and escarpment retreat driven by high order stream capture. The dashed line, indicating position of the escarpment in Frame 1, remains ﬁxed throughout as a reference point. (panel 1) Headward erosion progresses rapidly along an escarpment-orthogonal weakness (e.g., fracture) and approaches an initial capture point. (panel 2) Capture and reversal of escarpment-parallel drainage leads to rapid dissection and the progression of headward erosion along drainage-controlling weaknesses. Because of the rectilinear drainage network and low Upland relief, two additional captures are now imminent. A terrace may have been preserved by the beheading of the preexisting drainage. (panel 3) Capture of two more escarpment-orthogonal streams leads to rapid dissection of the entire basin affected by capture thus far. Rapid incision by the reenergized drainages isolates portions of the Upland and strands two new remnant terraces. (panel 4) Complete dissection of the captured basin leads to parallel escarpment and divide retreat. A terrace deposit remains at the now underfed headwaters of the far drainage. The near drainage, though beheaded, has been rejuvenated by an Upland base level drop that has locally steepened the terrain and scoured away any residual alluvium. The small ridge now located at the escarpment crest was once a local divide within the Upland interior, and the stream draining it (starred) has not been affected by divide migration although it now drains the main regional divide. Stream 29 may be representative of the ECD “overtaking” a subordinate, landward divide (Figs. 3, 5). Capture is not favored across this ridge or the more symmetric divide at the bottom of Frame 4, and this portion of the escarpment will enter a phase of stability. Headward erosion could proceed along another escarpment-orthogonal lineament to initiate a new cycle of capture and punctuated retreat of the far portion of the escarpment.

spatially isolated but occur in enough locations with sufficient indicative of more extensive weathering than that experienced by regularity to produce an overall parallel retreat over the lifetime of the other terraces. In contrast, orthogonal flow directions increase the the BRE. The long-term BRE retreat rate may therefore best be viewed potential for capture of high order channels and rapid adjustment of as a “weighted average” of local and short-lived rapid retreat events topography to the new divide location. Orthogonal drainages also interrupting long periods of quiescence where escarpment evolution facilitate capture across low-relief subordinate divides of the Upland is negligible. When these rare, localized but extremely rapid retreat (Fig. 10, panel 3), allowing basin-by-basin capture to propagate along rates are considered over the length of the feature, possibly for BRE strike. The impact of orthogonal flow directions and weak ~200Ma, a more reasonable “lifetime” parallel retreat rate is drainage-controlling structures on retreat rate is clearly seen where suggested. fluvial evidence of retreat is best preserved, such as in terraces E and N Our proposed retreat model is heavily dependent on a preexisting (Fig. 3A, B). Easily eroded fractures and lithologies permit rapid basin rectilinear drainage network on the Upland surface. Lineament dissection and speed the landward progression of headward erosion, analysis of areas of the escarpment in which alluvium is fresh and the source of subsequent captures further inland. divide asymmetry most pronounced indicate the role of orthogonal Considering the potential for repeated episodes of local but rapid drainage-controlling weaknesses in sustaining the retreat process retreat combined with extremely slow Upland lowering, interpreta- (Figs. 10, 11). Streams in the Appalachian Blue Ridge and Piedmont tion of the BRE as the final remnant of a Mesozoic rift-flank feature generally trend along two orientations: orogenic (and BRE) strike- may be reasonable. The initial BRE topography almost certainly parallel, controlled by weak lithologies or structures (e.g., Brevard formed well seaward of its present location, and quartz mylonite fault zone), and orogenic strike-orthogonal, controlled by fractures clasts in terraces B and E (Fig. 3A) indicate the Brevard/Bowens Creek and joints (Fig. 11). The enhanced erodibility of these drainage- fault zone was a drainage-controlling feature atop the Upland and did controlling features is fundamental to BRE evolution. Orthogonal flow not reactivate to initiate the BRE as suggested by White (1950). This directions allow for more frequent captures at high order points along idea is further supported by fluvial evidence of retreat atop the ECD Upland streams that facilitate a large increase in stream power and and BRE where it is located well seaward of the Brevard Zone in rapid dissection of the captured basin as illustrated by our conceptual western North Carolina (Fig. 3B). While the clasts we examined model (Fig. 10). Where flow directions are essentially opposite across represent only the most recent captures and retreat, they still suggest the divide, headward erosion captures insignificant Upland area and sufficient transport to restore parts of the escarpment near Triassic does not energize the capturing drainage, producing minimal slow basin border faults 70 km southeast. Whether the BRE initiated at retreat. Terrace I (Fig. 3A), preserved in such an area, contains clasts these structures or further southeast remains unknown. The present showing considerable pitting, breakage, and deep staining, all Piedmont drainage network suggests that the mechanism indicated P.S. Prince et al. / Geomorphology 123 (2010) 305–319 317

Terrace E

Terrace D

0 5 km N

Terrace N

0 5 km N

Fig. 11. Shaded relief maps of two areas, each shown with and without interpreted lineaments, showing the roughly orthogonal lineament trends prevalent in areas of the BRE that host well-rounded and fresh terrace alluvium (see Fig. 3 for location). The thickness of the interpreted lineament trends is a proxy for influence over the drainage network. Northeast-trending lineaments are associated with thrust faults and weak, sheared lithologies of the Brevard/Bowens Creek fault zone. Northwest-trending lineaments are joints and fractures that pass uninterrupted across the Brevard fault zone, suggesting post-Paleozoic origin. The rectilinear drainage network resulting from the interaction of these weak features facilitates repeated large capture events as well as rapid dissection of the captured basin (Fig. 10). Shaded relief images are captured from Google Maps, based on 10-m resolution topography. ©2009 Google-Map data ©2009 Google. by our study could have supported continued BRE retreat from an sample locations. As the high order capture events that ultimately initial location over 100 km to the southeast. Persistence of divide drive retreat are spatially isolated and transient because of the rapid asymmetry related to Upland stability (Sullivan et al., 2007) combines dissection of captured basins, escarpments may appear (and likely are) with active incision along the escarpment front (e.g., streams 19 and stable along most of their length at any given time (Fleming et al., ii; Fig. 3A) to suggest that the landform will continue to evolve and 1999; Cockburn et al., 2000; Sullivan et al., 2007; Vanacker et al., 2007). maintain its topographic youth despite an ~200 Ma erosional history. The active process may be difficult to observe unless cosmogenic Field evidence of the role of structurally controlled stream capture samples are taken in the immediate vicinity of a large knickpoint and punctuated retreat of the BRE complements studies of other formed by a recent capture event. Cosmogenic dating of captured passive margin escarpments. Our results illustrate the value of basins undergoing dissection, such as the Dan River Gorge (stream 19 applying more traditional methods of geologic study to constrain of Fig. 3A), will be an important way of testing the episodic retreat passive margin escarpment history. Alluvium has not yet been used as model and assessing its potential applicability to sinuous escarpments, a proxy for retreat of the Western Ghats, but the combination of such as southeast Australia, where gorge incision is known to outpace appropriate structure, variation in lithology, and apparent capture upland lowering (Nott et al., 1996). frequency (Harbor and Gunnell, 2007) suggests the likelihood of terraces preserved at or near its crest. The low relief, low erosion rates, 6. Conclusions preservation of soils, and rectilinear drainage network of the Sri Lankan Upland (von Blanckenburg et al., 2004; Vanacker et al., 2007), Fluvial terraces and beheaded stream valleys preserved atop the qualitatively similar to features of the Blue Ridge Upland, may have ECD at the BRE crest provide clear evidence of significant escarpment preserved evidence of parallel escarpment retreat driven by stream retreat associated with landward divide migration during the capture. The episodic retreat model highlights the importance of Cenozoic. Stream morphology and terrace location suggest this divide location and preexisting drainage network as controls over ongoing retreat process is driven by the episodic capture of large escarpment evolution (van der Beek et al., 2002). Our data also Upland drainages by steep streams flowing east to the Atlantic Ocean. indicate the importance of careful selection of cosmogenic erosion rate The structurally controlled rectilinear drainage network and low relief 318 P.S. Prince et al. / Geomorphology 123 (2010) 305–319 of the Upland facilitate repeated captures, which produce rapid Brown, R.W., Gallagher, K., Gleadow, A.J.W., Summerfield, M.A., 2000. Morphotectonic evolution on the south Atlantic margins of Africa and South America. In: dissection of the captured basin and localized landward migration of Summerfield, M.A. (Ed.), Geomorphology and Global Tectonics. John Wiley and the zone of maximum relief. When considered along the entire strike Sons, New York, pp. 255–281. length of the BRE, the combined effect of these local, but rapid, retreat Brown, R.W., Summerfield, M.A., Gleadow, A.J.W., 2002. Denudational history along a transect across the Drakensberg escarpment of southern Africa derived from events may lead to parallel divide and escarpment retreat over the apatite fission track thermochronology. Journal of Geophysical Research 107, long term. While the roundness of terrace clasts suggests many tens of 2,333–2,350. kilometers of transport, uncertain provenance and paleodrainage Cailleux, A., 1947. L'indice d'emoussé. Definition et premiere application. Compte- – networks preclude the use of this data in establishing a total retreat rendu sommaire de la Société Géologique de France 13, 250 252. Cockburn, H.A.P., Brown, R.W., Summerfield, M.A., Seidl, M.A., 2000. Quantifying passive distance or point of origin for the BRE. Nevertheless, the size and margin denudation and landscape development using a combined fission-track roundness of alluvium preserved at the BRE crest suggest extensive thermochronology and cosmogenic isotope analysis approach. Earth and Planetary – Upland drainage basins have been destroyed by escarpment and Science Letters 179, 429 435. Davis, W.M., 1902. Baselevel, grade, and peneplain. Journal of Geology 10, 77–111. divide retreat. Davis, W.M., 1903. The stream contest along the Blue Ridge. Geological Society of In contrast to recent numerical modeling and cosmogenic studies, Philadelphia Bulletin 3, 213–244. our results indicate that landward migration of the drainage divide Dietrich, R.V., 1957. Origin of the Blue Ridge escarpment directly southwest of Roanoke, – fi Virginia. Virginia Journal of Science 8, 233 246. produced by stream capture events can produce continued signi cant Dietrich, R.V., 1959. Geology and mineral resources of Floyd County of the Blue Ridge retreat of mature passive margin escarpments. The persistence of the upland, southwestern Virginia. Bulletin of Virginia Polytechnic Institute 52. retreat process long after the demise of initial rift topography may be Fleming, A., Summerfield,M.A.,Stone,J.O.,Fifield, L.K., Cresswell, R.G., 1999. Denudation rates for the southern Drakensberg escarpment, SE Africa, derived related to a combination of extremely slow Upland denudation and from in-situ-produced cosmogenic Cl-36: initial results. Journal of the Geological comparatively rapid escarpment retreat. The freshness of terrace Society 156, 209–212. alluvium and intact preservation of relict stream gradients suggest Flint, J.J., 1974. Stream gradient as a function of order, magnitude, and discharge. Water Resources Research 10, 969–973. that local retreat rates resulting from large captures may outpace Gallagher, K., Brown, R., 1997. The onshore record of passive margin evolution. Journal Upland lowering by three orders of magnitude. This differential of the Geological Society of London 154, 451–457. erosion maintains maximum divide asymmetry and the potential Gallagher, K., Hawkesworth, C.J., Mantovani, M.S.M., 1994. The denudation history of fi energy of Upland streams relative to the Atlantic base level, the onshore continental margin of SE Brazil inferred from apatite ssion track data. Journal of Geophysical Research, Solid Earth and Planets 99, 18,117–18,145. preserving the energetic driver of the retreat process. This model of Gasparini, N.M., Whipple, K.X., Bras, R.L., 2007. Predictions of steady-state and transient episodic retreat is consistent with areas of slow retreat along the BRE landscape morphology using sediment-flux-dependent river incision models. and other escarpments indicated by cosmogenic dating, but it also Journal of Geophysical Research, Solid Earth and Planets 112 (F3), F03S09. Gilchrist, A.R., Kooi, H., Beaumont, C., 1994. Post-Gondwana geomorphic evolution of predicts that very localized retreat rates in other areas could be at southwestern Africa — implications for the controls on landscape development least three orders of magnitude higher. Additional cosmogenic dating from observations and numerical experiments. Journal of Geophysical Research, along the BRE will be necessary to validate our model, and its Solid Earth 99, 12,211–12,228. Goede, A., 1975. Downstream changes in the pebble morphometry of the Tambo River, relevance to other passive margin escarpments is presently unclear. A eastern Victoria. Journal of Sedimentary Petrology 45, 704–718. single paradigm of passive margin escarpment evolution may not Goldrick, G., Bishop, P., 2007. Regional analysis of bedrock stream long profiles: exist, and these features may evolve through a range of mechanisms evaluation of Hack's SL form, and formulation and assessment of an alternative (the DS form). Earth Surface Processes and Landforms 32, 649–671. governed by drainage networks, divide location and asymmetry, Gunnell, Y., Harbor, D., 2008. Structural underprint and tectonic overprint in the structure, and lithology. In either case, the physical evidence of the Angavo (Madagascar) and Western Ghats (India)—implications for understanding role of divide migration in BRE retreat provides a useful additional scarp face morphology at passive margins. Journal of the Geological Society of India 71, 763–779 (Golden Jubilee). constraint on the evolution of this feature and may suggest a new Hack, J.T., 1957. Studies of Longitudinal Stream Profiles in Virginia and Maryland. U.S. source of data regarding the retreat mechanism of other passive Geological Survey Professional Paper. Report P 0294-B, pp. 47–97. margin escarpments. Hack, J.T., 1973. Drainage adjustment in the Appalachians. In: Morisawa, M. (Ed.), Fluvial Geomorphology. State University of New York, Binghamton, New York, pp. 51–69. Hancock, G.S., Kirwan, M.L., 2007. Summit erosion rates deduced from 10Be: – Acknowledgements implications for relief production in the Central Appalachians. Geology 35, 89 92. Harbor, D., 1996. Non-uniform erosion patterns in the Appalachian Mountains of Virginia. Geological Society of America Abstracts with Programs 28, 116. We wish to thank Editor-in-chief Dr. Richard Marston and three Harbor, D., Gunnell, Y., 2007. Along-strike escarpment heterogeneity of the Western anonymous reviewers whose comments have greatly enhanced this Ghats: a synthesis of drainage and topography using digital morphometric tools. Journal of the Geological Society of India 70, 411–426. manuscript. Hayes, C.W., Campbell, M.R., 1894. Geomorphology of the Southern Appalachians. National Geographic Magazine 6, 63–126. Heimsath, A.M., Chappell, J., Dietrich, W.E., Nishiizumi, K., Finkel, R.C., 2000. Soil production References on a retreating escarpment in southeastern Australia. Geology 28, 787–790. Hollerman, P., 1971. Zurungdungsmessen an Ablagerungen im Hochgebirge. Zeitschrift Bank, G., 2001. Testing the origins of the Blue Ridge escarpment. Master's Thesis, fur Geomorphologie 12, 205–237. Virginia Polytechnic Institute and State University, Blacksburg, Virginia. Hovermann, J., Poser, H., 1951. Morphometrische und morphologische Schotter-analysen. Battiau-Queney, Y., 1989. Constraints from deep crustal structure on long-term landform Proceedings of the 3rd. International Congress of Sedimentology. Groningen, The development of the British Isles and eastern United States. Geomorphology 2, 53–70. Netherlands, pp. 135–156. Bierman, P.R., Caffee, M., 2001. Steady state rates of rock surface erosion and sediment Hubbard, S.S., Coruh, C., Costain, J.K., 1991. Paleozoic and Grenvillian structures in the production across the hyperarid Namib desert and the Namibian escarpment, southern Appalachians: extended intrepretation of seismic reflection data. Tectonics southern Africa. American Journal of Science 301, 326–358. 10, 141–170. Bishop, P., Goldrick, G., 2000. Geomorphological evolution of the Eastern Australia Jenkins, R.E., Lachner, E.A., Schwartz, F.J., 1971. Fishes of the central Appalachian drainages: continental margin. In: Summerfield, M.A. (Ed.), Geomorphology and Global Their distribution and dispersal. In: Holt, P.C., Paterson, R.A., Hubbard, J.P. (Eds.), The Tectonics. John Wiley and Sons, New York, pp. 225–254. distributional history of the biota of the southern Appalachians part III: Vertebrates. Bohannon, R.G., Naeser, C.W., Schmidt, D.L., Zimmermann, R.A., 1989. The timing of Virginia Polytechnic Institute and State University, Blacksburg Virginia, pp. 43–117. uplift, volcanism, and rifting peripheral to the Red-Sea — a case for passive rifting. King, L.C., 1957. The uniformitarian nature of hillslopes. Transactions of the Edinburgh Journal of Geophysical Research, Solid Earth and Planets 94, 1,683–1,701. Geological Society 17, 81–102. Bowman, D., Shachnovich-Firtel, Y., Devora, S., 2007. Stream channel convexity induced King, L.C., 1962. The Morphology of the Earth: A Study and Synthesis of World Scenery. by continuous base level lowering, the Dead Sea, Israel. Geomorphology 92, 60–75. Hafner Publishing Company, New York. Braun, J., 2006. Recent advances and current problems in modelling surface processes Kirby, E., Whipple, K.X., 2001. Quantifying differential rock uplift rates via stream and their interactions with tectonics and crustal deformation. In: Buiter, S.J.H., profile analysis. Geology 29, 415–418. Schreurs, G. (Eds.), Analog and Numerical Modeling of Crustal-scale Processes: Kooi, H., Beaumont, C., 1994. Escarpment evolution on high-elevation rifted margins — Special Publication of the Geological Society of London, 253, pp. 307–325. insights derived from a surface processes model that combines diffusion, advection, Braun, J., van der Beek, P.A., 2004. Evolution of passive margin escarpments: what can and reaction. Journal of Geophysical Research, Solid Earth 99, 12,191–12,209. we learn from low-temperature thermochronology? Journal of Geophysical Kuenen, P.H., 1956. Experimental abrasion of pebbles. (2) Rolling by current. Journal of Research, Solid Earth and Planets 109, F04009. Geology 64, 336–368. P.S. Prince et al. / Geomorphology 123 (2010) 305–319 319

LaRue, J.-P., 2008. Effects of tectonics and lithology on long profiles of 16 rivers of the Seidl, M.A., Weissel, J.K., Pratson, L.F., 1996. The pattern and kinematics of escarpment Southern–Central Massif border between between the Aude and the Orb (France). retreat across the rifted continental margin of SE Australia. Basin Research 8, Geomorphology 93, 343–367. 301–316. Lindsey, D.A., Langer, W.H., van Gosen, B.S., 2007. Using pebble lithology and roundness Snyder, N.P., Whipple, K.X., Tucker, G.E., Merritts, D.J., 2000. Landscape response to tectonic to interpret gravel provenance in Piedmont fluvial systems of the Rocky Mountains, forcing: DEM analysis of stream profiles in the Mendocino triple junction region, USA. Sedimentary Geology 199, 223–232. Northern California. Geological Society of America Bulletin 112, 1,250–1,263. Matmon, A., Bierman, P.R., Enzel, Y., 2002. Pattern and tempo of great escarpment Spotila, J.A., Bank, G.C., Reiners, P.W., Naeser, C.W., Naeser, N.D., Henika, B.S., 2004. erosion. Geology 30, 1,135–1,138. Origin of the Blue Ridge escarpment along the passive margin of Eastern North McHone, J.G., 1996. Broad-terrane Jurassic flood basalts across northeastern North America. Basin Research 16, 41–63. America. Geology 24, 319–322. Steckler, M.S., Omar, G.I., 1994. Controls on erosional retreat of the uplifted flanks of the Mills, H.H., 1979. Downstream rounding of pebbles: a quantitative review. Journal of Gulf of Suez and northern Red Sea. Journal of Geophysical Research, Solid Earth and Sedimentary Petrology 49, 295–302. Planets 99, 12,159–12,173. Moore, A., Blenkinsop, T.G., 2006. Scarp retreat versus pinned drainage divide in the Sullivan, C., Bierman, P.R., Reusser, L., Pavich, M., Larsen, J., Finkel, R.C., 2007. formation of the Drakensberg escarpment, Southern Africa. South African Journal of Cosmogenic erosion rates and landscape evolution of the Blue Ridge escarpment, Geology 109, 599–610. southern Appalachian Mountains. Geological Society of America Abstracts with Moore, M.E., Gleadow, A.J.W., Lovering, J.F., 1986. Thermal evolution of rifted Programs 39, 512. continental margins — new evidence from fission tracks in basement apatites ten Brink, U., Stern, T., 1992. Rift flank uplifts and hinterland basins: comparison of the from southeastern Australia. Earth and Planetary Science Letters 78, 255–270. Transantarctic Mountains with the Great Escarpment of Southern Africa. Journal of Munro-Perry, P.M., 1990. Slope development in the Kerkspruit Valley, Orange Free Geophysical Research, Solid Earth and Planets 97, 569–585. State, South Africa. Zeitschrift fur Geomorphologie 34, 409–421. Tricart, J., Schaeffer, R., 1950. The study of erosion systems through the consideration of the Naeser, N.D., Naeser, C.W., Newell, W.L., Southworth, S., Weems, R.E., Edwards, L.E., “Roundness Index” of pebbles. Revue de Géomorphologie Dynamique 1, 151–1,152. 2006. Provenance studies in the Atlantic coastal plain: what fission track ages of Tucker, G.E., Slingerland, R.L., 1994. Erosional dynamics, flexural isostasy, and long- detrital zircons can tell us about the erosion history of the Appalachians. Geological lived escarpments. A numerical modeling study. Journal of Geophysical Research, Society of America Abstracts with Programs 38, 503. Solid Earth and Planets 99, 12,229–12,242. Nott, J., Young, R., McDougall, I., 1996. Wearing down, wearing back, and gorge extension van der Beek, P.A., Braun, J., 1999. Controls on post-mid-Cretaceous landscape evolution in the long-term denudation of a highland mass: quantitative evidence from the in the Southeastern Highlands of Australia: insights from numerical surface process Shoalhaven catchment, southeast Australia. Journal of Geology 104, 224–232. models. Journal of Geophysical Research, Solid Earth and Planets 104, 4945–4966. Ollier, C.D., 1984. Morphotectonics of continental margins with great escarpments. In: van der Beek, P.A., Summerfield, M.A., Braun, J., Brown, R.W., Fleming, A., 2002. Morisawa, M., Hack, J.T. (Eds.), Tectonic Geomorphology. Allen and Unwin, London, Modeling post-breakup landscape development and denudation history across the pp. 3–25. southeast African (Drakensberg escarpment) margin. Journal of Geophysical Partridge, T.C., 1998. Of diamonds, dinosaurs, and diastrophism: 150 million years of Research, Solid Earth and Planets 107 (B12), 2351. landscape evolution in southern Africa. South African Journal of Geology 101, Vanacker, V., von Blanckenburg, F., Hewawasam, T., Kubik, P.W., 2007. Constraining 167–184. landscape development of the Sri Lankan escarpment with cosmogenic nuclides in Partridge, T.C., Maud, R.R., 1987. Geomorphic evolution of southern Africa since the river sediment. Earth and Planetary Science Letters 253, 402–414. Mesozoic. South African Journal of Geology 90, 179–208. Virginia Division of Mineral Resources, 2003. Digital representation of the 1993 Pazzaglia, F.J., Brandon, M.T., 1996. Macrogeomorphic evolution of the post-Triassic geologic map of Virginia. Virginia Division of Mineral Resources Publication 174 Appalachian Mountains determined by deconvolution of the offshore basin [CD-ROM; 2003, December 31]. Adapted from Virginia Division of Mineral sedimentary record. Basin Research 8, 243–254. Resources, 1993, Geologic map of Virginia. Virginia Division of Mineral Resources, Pazzaglia, F.J., Gardner, T.W., 2000. Late Cenozoic landscape evolution of the US Atlantic scale 1:500,000. passive margin: insights into a North American great escarpment. In: Summerfield, von Blanckenburg, F., Hewawasam, T., Kubik, P.W., 2004. Cosmogenic nuclide evidence M.A. (Ed.), Geomorphology and Global Tectonics. John Wiley and Sons, New York, for low weathering and denudation in the wet, tropical highlands of Sri Lanka. pp. 282–302. Journal of Geophysical Research, Solid Earth and Planets 109, 1–22. Penck, W., 1953. Morphological Analysis of Landforms. (translation by Hella Czeck and Ward, D.J., Spotila, J.A., Hancock, G.S., Galbraith, J.M., 2005. New contstraints on the late K.C. Boswell) St. Martin's Press, New York. Cenozoic incision history of the New River, Virginia. Geomorphology 72, 54–72. Persano, C., Stuart, F.M., Bishop, P., Barfod, D.N., 2002. Apatite (U–Th)/He age Weissel, J.K., Seidl, M.A., 1998. Inland propagation of erosional escarpments and river profile constraints on the development of the Great Escarpment on the southeastern evolution across the Southeastern Australia passive continental margin. In: Wohl, K.T.E. Australian passive margin. Earth and Planetary Science Letters 200, 79–90. (Ed.), Rivers over Rock: Fluvial Processes in Bedrock Channels: Geophysical Monograph, Pique, A., Laville, E., 1995. L'ouverture initiale de l'Atlantique central. Bulletin de la 107. American Geophysical Union, Washington, D.C., pp. 189–206. Societe Geologique de France 166, 725–738. Whipple, K.X., 2004. Bedrock rivers and the geomorphology of active orogens. Annual Poag, W.C., Sevon, W.D., 1989. A record of Appalachian denudation in postrift Mesozoic Review of Earth and Planetary Sciences 32, 151–185. and Cenozoic sedimentary deposits of the U.S. Middle Atlantic continental margin. White, W.A., 1950. Blue Ridge Front — a fault scarp. Geological Society of America Geomorphology 2, 119–157. Bulletin 61, 1,309–1,346. Pratt, T.L., Coruh, C., Costain, J.K., Glover, L., 1988. A geophysical study of the earth's Widdowson, M., Gunnell, Y., 1999. Lateritization, geomorphology and geodynamics of a crust in central Virginia: implications for Appalachian crustal structure. Journal of passive continental margin: the Konka and Kanaran lowlands of western peninsular Geophysical Research, Solid Earth and Planets 93, 6,649–6,667. India. In: Thiry, M., Simon-Coincon, R. (Eds.), Palaeoweathering, Palaeosurfaces, and Roe, G.H., Montgomery, D.R., Hallett, B., 2002. Effects of orographic precipitation Related Continental Deposits 27. : International Association of Sedimentologists variations on the concavity of steady-state river profiles. Geology 30, 143–146. Special Publication, 28. Wiley-Blackwell, New York, pp. 245–274. Rust, D.A., Summerfield, M.A., 1990. Isopach and borehole data as indicators of rifted Wright, F.J., 1927. The Blue Ridge of southern Virginia and western North Carolina. margin evolution in southwestern Africa. Marine and Petroleum Geology 7, 277–287. Journal of the Scientific Laboratories, Denison University 22, 116–132. Sadler, P.M., Reeder, W.A., 1983. Upper Cenozoic, quartzite-bearing gravels of the San Young, R., McDougall, I., 1993. Long-term landscape evolution: Early Miocene and Bernardino Mountains, Southern California; recycling and mixing as a result of modern rivers in southern New South Wales, Australia. Journal of Geology 101, transpressional uplift. In: Andersen, D.W., Rymer, M.J. (Eds.), Tectonics and 35–49. Sedimentation along the Faults of the San Andreas System. Society of Economic Paleontologists and Mineralogists, Pacific Section, Los Angeles, pp. 45–57.